CHAPTER SEVEN

PRACTICE: D&T + D&I

If there is no struggle, there is no progress.

—FREDERICK DOUGLASS

In this final chapter, I elaborate on the path to creating effective diversity and inclusion programs, programs that produce diversity bonuses. This is not an easy task. Once, in presenting the logic of diversity bonuses to NASA employees, I commented that social science was harder than rocket science because social science must understand people, and people are an unpredictable, diverse lot.

The difficulty of creating an effective diversity and inclusion program should be apparent to nearly everyone. Each year, corporate America spends billions on diversity and inclusion. In 2015, Google planned on spending $150 million on diversity initiatives, and Apple spent $50 million. A preponderance of evidence suggests that all of this spending barely budges the needle on numbers and that efforts to reduce discrimination often result in backlash or diversity fatigue.¹

At this point, a quick summary of what we know, and what it appears we have to do, goes as follows: On a variety of tasks, diverse teams have the potential to produce exceptional performance; that is, better performance than individuals or homogeneous groups. As a rule, on complex tasks, the best team will not consist of the best individual performers. The best team will instead consist of people with diverse, relevant repertoires.

It then follows that organizations should apply their best judgment, informed by theory and data, to create diverse, high-potential teams; that they should also create a culture of trust and a shared sense of mission; and that they should design practices and procedures to achieve bonuses. Given that diverse teams add costs in effort and time, diversity should be applied to high-value, complex tasks. The easy, low-margin stuff can be left to individuals. The goal is obvious: inclusive teams with germane diversity on the hard high-value problems, strategic diversity, and inclusion on complex tasks.

Attaining that goal requires effort and practice. To arrange different thinkers wrapped in diverse identities around a conference table and expect a bonus is to engage in the type of magical thinking that has been the Achilles’ heel of diversity initiatives. Success does not depend on how diverse the people in the room look; it depends on the team’s ability to leverage their diverse repertoires. Tossing diverse people in a room and asking them to be innovative will more often produce a dog’s breakfast than an iPhone or a breakthrough app like Pokemon Go.

At a minimum, any successful diversity and inclusion policy must include six Ms. Organizations must message from the top and link diversity to mission. The organization must manage teams with the goal of achieving diversity bonuses. That means creating an inclusive culture. It also means applying germane diversity on complex high-value opportunities. Organizations must also measure performance and provide mentors for underrepresented employees. And, last, organizations must tie inclusive behaviors to merit; that is, they must reward people who advance diversity and inclusion.

These six Ms work in concert with one another. A CEO who presents glorious imagery of how differences lead to better outcomes has greater impact if she also puts in place policies and norms that enable bonuses to arise—that is, if she manages to produce bonuses.

These six Ms must be taken seriously. Among the companies I have visited, I have been struck by the correlation between success in diversity initiatives and buttoned-down, effective management practices on the other aspects of their businesses.

How any given organization builds good practices depends on the challenges they face, their current employees, and their employee pool. There can be no one-size-fits-all solution. There do exist a few areas where almost all organizations need to improve: bias reduction, the creation of inclusive cultures, and developing people and team analytics. First, though, I will talk about practice.

WE ARE TALKING ABOUT PRACTICE

In The Difference, I wrote of the parable of the bikes. I asked readers to imagine asking a group of five- or six-year-old children to run as far as they can in ten seconds and then to place those same kids on bikes and redo the experiment. I noted that while the average distance traveled in the two cases would be about the same, the very best performers would be children on bikes, and so would the worst.

Homogeneous groups perform like the kids who run. They produce the B-pluses. They rarely do anything amazing. Identity and cognitively diverse teams of people perform like the children on bikes. They produce exceptional results. We saw that in the data on academic papers and teams. Look into most great success and you find diversity. For example, the creative team behind the record-breaking musical Hamilton includes Lin-Manuel Miranda, a New Yorker of Puerto Rican descent, Jeffrey Seller, a gay, white, male Jewish adoptee from Detroit, and Jill Furman, who grew up in a wealthy Manhattan family.

But diverse groups, like the children on the bikes, can also fail miserably.

Creating successful diverse teams and inclusive organizations requires practice. That practice is made easier when we know that bonuses exist. Permit me an extended analogy. In 1979, the National Basketball Association introduced the three-point shot. Teams that could make the three-point shot earned a bonus: three-point shots were worth 50 percent more than normal shots. No one needed to write an entire book explaining the three-point bonus.

What was needed was practice. Though the best players in the world by most metrics, NBA players lacked the ability to make the longer shots. In that first year, NBA teams took on average three shots per game from behind the arc, making a paltry 28 percent. Multiplying that percentage by 1.5 corresponds to 42 percent for a two-point shot. In 1979 NBA players made two-pointers at a 48 percent clip. The three-point bonus went unrealized.

Practice took time. Two years after the introduction of the three-point line, the 1981–1982 NBA champions, the Los Angeles Lakers, starred Magic Johnson, James Worthy, and Kareem Abdul-Jabbar, three of the greatest players in the history of the NBA. They made only thirteen three-pointers out of ninety-four attempts. That translates to about 14 percent. Those Lakers did not realize a three-point bonus. To their credit, they did not even try for one.

Had NBA players relied only on empirical analysis, the three-pointer would have remained a minor part of the game. It was not producing any bonuses. And yet, every general manager, coach, and player saw the potential 50 percent bonus. That led to practice. If we jump ahead to 2015–2016, we see the residue of that effort. In 2015–2016, NBA teams averaged more than twenty three-point shots per game. In a single playoff game in 2016, the Cleveland Cavaliers torched the Atlanta Hawks with twenty-five three-pointers, a number that nearly doubles what the 1981–1982 Lakers made over the entire season.

Three-pointers now produce a bonus as well. League-wide, players connect on over 35 percent of three-point attempts. This equates to more than 50 percent accuracy on two-point shots. In addition, the very best teams shoot more three-pointers on average and make them with higher accuracy.

Knowledge of the three-point bonus provided a rudder and compass. Players spent hours in the gym—or, in the case of 2015 and 2016 league MVP Stephen Curry, on a dirt driveway—firing from long range. In 2016, Curry made more than four hundred three-pointers and connected at a rate above 45 percent.

Organizations need practice as well. They need to learn how to compose teams, and how to engage in behaviors that produce the bonuses. And they need data and theory to guide practices. Unlike basketball skills, which can be honed individually, the skills needed to produce diversity bonuses require participation in teams.

Practice, and therefore inclusion, entails cost. Schools, colleges, and universities can allay some of that cost by providing environments for young people to learn behaviors that produce diversity bonuses. Failure in the classroom—a botched homework assignment—won’t lower anyone’s profits or harm the environment. And, as might be expected, studies show that students who engage in diverse groups at school encounter more ideas and see the value of inclusivity. They continue to seek diverse groups. As adults, they are more likely to live in and work in integrated communities.²

BIAS REDUCTION

To learn behaviors that produce bonuses within diverse teams requires that the organization include diverse people. Our workplaces must be diverse, so must be our universities. Efforts to create diverse workforces must overcome biases that disadvantage women and members of minority groups. This is step one.

To get diverse people on the bus, so to speak, diversity and inclusion training programs often begin (and sometimes end) with bias-reduction training. Reducing discrimination is the low fruit in the business case for diversity as these programs receive little pushback from employees. My experience has been that this is true even for people who reject normative arguments based on historical and current discrimination.

People who work at Boeing want to hire the most talented engineers. The employees at PIMCO want to hire the smartest market analysts. Everyone prefers to choose from a larger pool of talent. That larger pool of talent also consists of a more cognitively diverse pool of talent. That is true even if identity diversity does not correlate with relevant cognitive diversity.³

Bias training is needed because even though the law prevents discrimination based on (some) identity characteristics, discrimination persists. As Thurgood Marshall noted in a 1988 speech to the American Bar Association, “A child born to a Black mother in a state like Mississippi … has exactly the same rights as a white baby born to the wealthiest person in the United States.” He then added, “It’s not true.”⁴

He was correct in 1988. He would also be correct now.⁵ Evidence shows that discrimination occurs at multiple layers across a variety of contexts. Women and applicants with African American–sounding names receive fewer job callbacks, lower salary offers, and lower competency rankings, despite identical resumes.⁶

Acts of bias can be unintended, and even unconscious.⁷ When making hiring decisions, people err in the direction of homogeneity. We falsely believe that people who share our identity are smarter and more capable. We are not as connected across identity groups as we need to be.⁸ Even if people did not prefer their own group, discrimination could still arise if people could more accurately evaluate people from their own identity group.⁹ For these reasons, unconscious-bias training has become a staple of diversity training.

Nepotism and social network effects also contribute to bias. People pull strings for their children and their friends’ children. They help get them summer internships. They arrange college interviews. These people’s friends and their children tend to belong to the same identity groups and have similar experiences. Both reduce cognitive diversity.

Last, as mentioned in the previous chapter, many companies apply a common rubric to applicants. This can also produce bias. If members of an identity group enter college less prepared, they may earn lower grades. Or they might not go to the “right” school or have the same opportunities.

The full list of biases—direct, unintended, unconscious, social network based, and rubric based—combine to disadvantage some groups. Attracting and maintaining a diverse pool of employees requires not just reducing them but eradicating them completely. Getting close is not good enough. Small biases accumulate to form large biases. That is not a moral prescriptive. It is based on the mathematics of bias accumulation.

The Accumulation of Bias

That mathematics of bias accumulation can be best understood through an example. As a first step, we need to calibrate the magnitude of racial and gender biases. Single-study biases range from 5 percent to 25 percent for salary offers.¹⁰ Similar-size biases seem to exist in other contexts. Manipulating the race of the seller of a baseball card from white to black lowers the price on eBay 20 percent.¹¹

These one-shot measures reveal small biases. I believe that to be accurate. Bias exists but not at horrible levels. However, these measures understate the accumulated effects of biased decisions. A person’s career consists of dozens if not hundreds of evaluations and perhaps dozens of opportunities for success.

Suppose that to make senior partner at a law firm or to reach the executive suite in a Fortune 500 company, a person must pass ten hurdles. These could be performing well in an interview, attracting a new client, or excelling in a leadership role. Each hurdle is an up or out, a win or lose.

A person with a 50 percent chance of success at each juncture stands a one-in-a-thousand chance of passing all ten and reaching the C suite. If we assume women and minorities are 10 percent less likely to earn a promotion, so they pass 40 percent of the time, then they reach the C suite at a rate of one in ten thousand. That’s one-tenth the chance. In a ten-level hierarchy, 10 percent discrimination produces 90 percent fewer women and minorities at the top.

Bias training counters these pernicious accumulated effects of small biases. It can increase awareness so that people ask themselves, Am I selecting this candidate or taking a risk on this candidate because she looks like me, or because she’s good? Bias training can also involve instituting changes in process. Process solutions include requirements to interview minority candidates, mandating at least one woman or minority on each panel or committee, or requiring more than one evaluation rubric.¹²

Eradicating bias is not as simple as it might seem. A university postdoctoral hiring committee might disproportionately hire faculty who attended elite graduate programs where students are thought to receive better training and to have cleared a higher admissions hurdle. A candidate who excelled at Thomas Jefferson High School for Science and Technology and Yale College, and then earned a PhD at Caltech, is perceived to be better coached by stronger scholars than someone who attended Middleville Thornapple-Kellogg High School in rural Michigan and Ferris State College, and earned a PhD at Western Michigan University.¹³

The committee should neither ignore the second candidate nor feel compelled to hire him under an inclusion initiative. Instead, the committee has to do the hard work of looking at the candidates’ repertoires as well as their accomplishments. If the university already employs two postdocs from the same lab that produced the first candidate, they might well give the second candidate a much longer look.

INCLUSIVE CULTURES

Once a company has diverse people in the room, those people must act to produce bonuses; that is, they must act inclusively. Creating an inclusive culture demands more than listing inclusivity as a core value and rewriting a few sentences on www.ourcompany.com. Meaningful inclusion requires a more organic bottom-up process guided by managers who shape behavior, motivate and inspire employees, guide actions, and create meaning.

Adding inclusion as a core value is still a good thing, and hundreds, if not thousands, of companies ranging from U.S. Steel to Cold Stone Creamery to Better Homes and Gardens Real Estate do so. Better Homes and Gardens Real Estate replaced innovation with inclusion to encourage employees to become better listeners and to be more open to new ideas. Ironically, that change could make the firm more innovative.

Inclusion can take a variety of forms. Here, what I mean by an inclusive culture is one in which people have the ability to apply their full repertoires. A lack of inclusion means that someone feels that she has something to add and does not or cannot. An inclusive culture need not involve people sharing their personal narratives and beginning each sentence with the phrase, “I appreciate and respect your position.” That said, people will be more willing to share ideas if they feel safe, respected, and validated.

Recall that if people share ideas, a team can be as good as its best member, so better than the average. If, in addition, ideas are challenged and combined, they can be better than those of their best member. Those challenges and deep engagements with ideas are a necessary component of the type of inclusive culture that maximizes bonuses.

Kim Scott, a management consultant, refers to the practice of challenging and improving as “exercising radical candor.” She sees this as a necessary behavior for a successful manager. Scott categorizes the behavior of managers on two dimensions: caring personally and challenging directly. Managers who care personally but do not challenge directly fall into the ruinous empathy box. To practice radical candor, a boss must be both caring and challenging. Caring gets the diverse idea into the room. Challenging improves it.¹⁴

The investment firm Bridgewater and Associates relies on an extreme form of radical candor that founder Ray Dalio refers to as a meritocracy of ideas. The firm’s culture prizes openness, transparency, and honesty above all. Any and all ideas put forward face stiff challenges. Dalio believes that his principles best advance Bridgewater’s core mission of understanding how the world works. Their form of inclusivity may not be for everyone, but it does achieve diversity bonuses.

Google’s lessons learned led them to similar policies. They too value dissent and encourage transparency. Dissent, by the way, is also a core value of McKinsey and Intel. Google also relies heavily on teams. Teams hire, promote, assign salaries, and even fire employees. Transparency, dissent, and teams: each increases the likelihood of good actions. Google also seeks talent. They want smart people.¹⁵

At Google, inclusion also means freedom. In their 2004 initial public offering letter, Larry Page and Sergey Brin described how employees could devote 20 percent of their time to new ideas and projects. Employees could form impromptu teams to leverage diverse talents to create new products and services. The list of in-house successes includes both Google News and Gmail. In practice, not every employee uses her full 20 percent. Less important than the actual percentage of time spent on individual projects is the fact that any employee could allocate one day a week to an idea and could build a team of Googlers to pursue it further. That possibility epitomizes an inclusive, flexible culture.

These types of inclusive practices promote diversity bonuses. That does not mean that they are ideal for every organization. The Nuclear Regulatory Commission does not want its employees experimenting with the reactor. It wants its employees to follow protocols. As a rule of thumb, an organization that depends on stability and control may find that more hierarchical, less inclusive cultures lead to better performance.¹⁶

Inclusive cultures make sense for companies that operate in fast-changing, that is, complex, environments. Thus, the trend away from routine work and toward nonroutine cognitive work spurs organizations to promote inclusion. The stories of Katherine Johnson and Dorothy Vaughan from Hidden Figures again prove illustrative. The advent of computers made hand calculations obsolete. Katherine Johnson became part of a team who applied techniques from analytic geometry to determine the calculations the computers would make. Dorothy Vaughan became a supervisor of Fortran computer programmers. Their transition to nonroutine work helped put people on the moon.

Thus, the trend toward cognitive nonroutine work in the corporate world creates pressure for more inclusive cultures. That trend occurs within organizations as well. A half century ago, Bell Canada made the bulk of its profits from providing phone service. Quality control and the maintenance of service were essential to their business success. Beginning in 1983, Bell Canada began diversifying. Now rebranded as the conglomerate BCE, it competes in a range of markets including wireless, media, and sports. Those markets demand flexibility. BCE CEO George Cope’s corporate profile lists his ability to build high-performance teams as one of two primary strengths.¹⁷

PEOPLE OR TEAM ANALYTICS

The bleeding edge of diversity and inclusion programs rely on cutting-edge analytics. They leverage data to create the best teams. Identifying criteria or attributes of successful groups represents the frontier.

At the moment, almost all hiring practices suffer from a common error. Organizations evaluate people as individuals when those people will work in teams. Some corporate human resource departments and graduate student admissions committees assign scores to grades, letters of recommendation, and so on to produce a cumulative score. Those applicants with the highest score get hired or admitted.

To ensure talent, organizations demand that people satisfy certain criteria to get a job. According to legend, Google’s hiring of Vint Cerf, an Internet pioneer who has been awarded the National Medal of Technology, the Turing Award, the Presidential Medal of Freedom, and the Marconi Prize; who has been elected to the National Academy of Engineering; and who has received more than twenty honorary degrees, was delayed because he did not submit an undergraduate transcript.

These practices violate the no-test results discussed in this book. On complex tasks, no single test applied to individuals can identify the best team. Some organizations continue to hire based on grade point averages despite the fact that the organization with probably the most data, Google, has evidence that it does not work. Laszlo Bock has described grade point averages and IQs as far less important than problem-solving ability in predicting success at Google.¹⁸

The growing field of people analytics seeks to identify characteristics of successful teams. What is known so far is that good managers play a role, as does the ability of team members to read the emotions of others.¹⁹ The aforementioned research on the millions of academic papers and patents implies that strategic team choice works. When academics form teams, they do not walk down the hallway and look for a random smart person. They search their contacts or the Internet for someone with a cognitive repertoire that complements their own. We know that, because the best papers combine diverse, deep references.²⁰

If a recipe for success consists of two parts—getting the right people in the room (diversity) and creating the right space for them to produce bonuses (inclusion)—then theory and data can assist us in both parts. Suppose a firm has to make a prediction about the effects of a change in the regulatory environment. Rather than have one person idiosyncratically select a diverse team, a team might be queried to generate ideas about what types of diversity might be relevant. What perspectives, knowledge, and tools might be useful in making an accurate prediction?

If possible, past data could be leveraged. I once visited a company that could trace its success to approximately one hundred up-or-down decisions on whether to advance products down the pipeline. When I asked what the data showed, that is, which people were good at making predictions, who made correlated predictions, and whether the company had evidence that certain pairs of people were never both wrong, the company replied that it did not keep data on people’s predictions.

Nor did another firm that relied heavily on in-house evaluations of product features. That firm, which has a market capitalization in excess of $100 billion, did keep data on employees’ NCAA basketball pool predictions and was able to tell me that the entry that averaged the selections of everyone was in the top three of over a thousand participants in that year’s contest. That success should have led them to average feature evaluations as well.

At both of these companies, the forecasts drive decisions that put tens or hundreds of millions of dollars at stake. To make those decisions, they relied on diverse teams, but they made no scientific effort to build optimal teams. They relied on seat-of-the-pants thinking. If Google’s former People Team leader Lazlo Bock were consulting for either of those companies, he would run regressions on the efficacy of decisions and team composition. If Bridgewater’s Ray Dalio were running either company, he would go back to the videotape any time an error was made. He would see if people ignored the correct idea or, even worse, if the right idea was not in the room. If the latter, he would look to hire someone whose background would make her likely to have the idea. If Phil Tetlock or Barb Mellers were in charge, they would identify diverse superforecasters and then combine them into a team.

CONCLUSION

Progress toward creating productive inclusion requires practice. The formal models reveal how cognitive diversity can contribute on a variety of tasks. They show how groups and teams whose members possess diverse representations, models, knowledge, and heuristics can make more accurate forecasts, find better solutions to problems, come up with more creative ideas, provide broader and deeper evaluations of policies and strategies, and better discern what is true.

The logic shows that diversity bonuses occur primarily on complex tasks that involve multiple dimensions or variables. Implicit in the claims are assumptions about each participant’s willingness and ability to contribute and the potential to contribute across diverse representations and models. We should therefore expect to put in effort if we want to achieve diversity bonuses. Ample empirical evidence shows that bonuses can be achieved: the best academic research, the most innovative patents, and the best investment decisions are all done by teams. And those teams are cognitively diverse and often identity diverse as well.

The potential for diversity bonuses offers an alternative frame. A focus on equity alone leads to a fixed-sum mind-set. We think in terms of tradeoffs. Diversity-bonus thinking enables us to see how our differences can make us more innovative, resilient, and prosperous. It points to how we might enlarge the pie instead of negotiating over the sizes of our current slices.

Progress requires a specific type of practice. We must rid ourselves and our organizations of conscious, unconscious, and structural biases. We must create environments in which everyone has an opportunity to contribute.

Herein lies one of the core challenges in creating effective diverse teams: success requires unity and difference. Successful diverse teams must be united in their goals. At the same time, team members must appreciate, encourage, and engage their differences. They cannot check their identities at the door. They must bring their whole selves—their identities, their experiences, their education and training—to achieve bonuses.

To conclude, I have shown logic and evidence for how germane cognitive diversity produces bonuses on the high-value complex tasks that predominate in our modern economy. I have shown how not any diversity will do. The diversity must be germane to the task. In some cases, differences in education or training may be most relevant. In others, our identities or experiences will have larger effects.

The theory and evidence demonstrates the need for a purposeful and strategic mind-set to group composition, hiring policies, and organizational culture and practices if we hope to produce diversity bonuses. The path will not be easy. Achieving these bonuses is a complex task. There may be few one-size-fits-all solutions. What works for Microsoft may not work for Louisiana Tech. What works for IBM may not work for Disney. Within each organization, thoughtful, diverse teams will be needed to think through how to identify and tap into diverse talent and how to create environments within which all people contribute and thrive.

I end on a hopeful note. Evidence of diversity bonuses becomes more compelling day by day, week by week. We appear to be getting better at creating effective diverse teams. Best of all, the people who belong to the teams and groups that produce diversity bonuses see the synergy between the normative ideal of an integrated, inclusive society and the economic ideal of an optimal team. They see inclusion as necessary to our collective success. May we all reach that place soon.